Units per hour: the labor number your cost report cannot give you

Your labor cost report says you are at seventy percent of the labor budget on a job that is not finished.

That sentence is either very good news or the worst news of the quarter, and the report cannot tell you which. Seventy percent of the budget with eighty percent of the work installed is a job you will make money on. Seventy percent with sixty percent installed is a loss that is already baked in, and you have five weeks left to notice.

The number that separates those two jobs is a ratio: how much work your crew puts in per hour. Many subcontractors never track it, because the accounting system has no place to put the top half of the fraction. This article is about building that number anyway, out of things your office already has.

Dollars tell you where you are. Units per hour tells you where you are going.

A cost report is a rear-view mirror with no distance markings. It compares money spent to money budgeted, and it is silent on the only question that matters mid-job: at the rate we are actually going, what will this cost when it is done?

Units per hour answers that, because it extrapolates. If your estimate assumed a crew would set fourteen devices per labor hour and the last three weeks have run at nine, you do not need a forecast model. You need a multiplier. The remaining work will take roughly one and a half times the hours left in the budget, and that is a number you can take into a conversation this week instead of a write-down in March.

It also survives a change in crew size, which dollars do not. Adding two people to a job makes the cost curve steeper and tells you nothing by itself. If the ratio holds when you add them, the scope was simply bigger than you thought. If the ratio drops, you have just paid more money to go slower, which happens more often than anyone admits.

Pick the unit your trade already counts

The unit has to be something a foreman can count without being asked twice. If measuring it takes a tape, a takeoff and half an hour, it will be measured once and then never again.

Most trades already have one:

• Electrical. Devices set, fixtures hung, terminations made, feet of conduit run by size, panels trimmed out.
• Mechanical and HVAC. Pounds or linear feet of duct installed by size class, pieces of equipment set, feet of pipe by diameter, diffusers and grilles installed.
• Plumbing. Fixtures roughed, fixtures trim-set, feet of pipe by size and material, floor drains and carriers set.
• Fire sprinkler. Heads installed, feet of main and branch line, drops, risers.
• Drywall and framing. Linear feet of track and stud, boards hung, square feet taped to a finish level, openings framed.
• Glazing. Lites set, linear feet of frame, storefront bays, doors hung.
• Concrete. Cubic yards placed, square feet of forms set and stripped, square feet finished.

Two rules about the unit itself. One unit per phase, not one per job — “square feet of building” is not a unit, because rough-in, trim and finish are three different crews doing three different things at three different rates, and averaging them hides all three. And count what is installed, not what is delivered. Material on the deck is not production, and treating it as production is the most common way a sub talks itself into believing a slipping job is on track.

The denominator is where everybody cheats

The hours are the easy half to get and the easy half to fudge, because every hour you quietly leave out makes the ratio look better.

Four rules keep it honest.

Only the hours spent on that scope count. If three of your six people spent Tuesday moving material for the GC, those hours do not belong in the install ratio. They belong somewhere — a general-conditions code, a back-charge, or a ticket — but not in the denominator of the work they did not do.

Extra-work hours come out. Hours spent on a change order or a T&M ticket are not base-contract production, and leaving them in makes your base scope look slow while the extra work looks free. This is the most expensive version of the mistake, because it costs you the bid data too.

The apprentice counts. It is tempting to measure the ratio on your journeymen and treat the helper’s hours as overhead. Do not. The ratio you need for the next bid is the ratio of your real crew mix, including the person who is learning. If your mix changes, the ratio should move — that is information, not noise.

The bad days stay in. A ratio that excludes the week it rained, the week the hoist was down and the week two people were out sick describes a job you have never run. Keep those weeks, and keep the reason next to them in the daily log, because the reason is what turns a bad number into a claim instead of an embarrassment.

Getting hours attributed this cleanly is a collection problem, not a reporting problem, and it is the subject of its own article: tracking labor hours by job and cost code. The short version is that the hour has to be recorded once, by the person who was there, on the day, against the part of the job it was spent on.

Build the baseline from the job you just finished

You do not need a published production rate book, and if you have one, the rates in it were measured on somebody else’s crew in somebody else’s building.

You need three numbers per phase:

1. The rate your estimate assumed. Divide the estimated quantity by the estimated hours. Most estimators never write this down as a ratio, but it is in there, and it is the number your price was built on.
2. The rate you actually achieved last time. Same arithmetic, from a finished job’s final quantities and final hours. The first time you do this you will find a gap between one and two, and that gap is your estimating error — which is worth more than any single job’s data.
3. The rate this job is running at right now. Rolling, two or three weeks, by phase and by area.

Three jobs’ worth of number two is enough to bid with. One job’s worth is enough to argue with.

Read it weekly, by area, and expect the first floor to be slow

Monthly reporting kills this metric. A month is long enough for the crew to finish an area, start the next one and change the mix, and the average of all that is a number no decision can be made from. Weekly, by area, is the grain where the ratio can still change an outcome.

Expect the curve to be ugly at the start. The first area of any repetitive scope runs slow — layout is being figured out, the stocking plan is wrong, the detail nobody read is being discovered in the field. That is the learning curve, and if you treat the first area’s rate as the job’s rate you will panic in week two of a job that is fine. What you watch instead is the shape: the rate should improve through the second and third areas and then flatten. A rate that never improves means the problem is not learning. A rate that improves and then falls back means something changed on site.

A bad ratio has five causes, and only two of them are your crew

This is the part that makes the number useful rather than demoralizing. When the ratio drops, work the list in this order.

Out-of-sequence or stacked work. You are in the area before the predecessor trade finished, or four trades are in a corridor built for one. Your people are working, moving and working again. This is often the biggest cause, and it is not yours — but it is only recoverable if the logs recorded it at the time. See documenting a delay.

Rework. Something was installed, something else changed, it came back out. If rework has no code of its own it is invisible, and it gets reported to you as your crew being slow.

Scope creep nobody billed. The extra hangers, the extra firestopping, the fixture relocations done as a favor. Each one is small, none of them was priced, and together they are your margin. If the ratio drops and nobody can name a delay, look here first.

Wrong crew size for the work face. Too many people in too little area, or a crew that spends half its day staging because it is too small to split. Both show up as a labor problem and are really a planning problem.

Overtime decay. The tenth hour of a ten-hour day and the sixth day of a six-day week often do not produce like the first ones, and they cost more. If the schedule recovery plan is overtime, the ratio is how you find out what you actually bought.

Three ways the number pays for itself

The next bid. This is the big one. A sub who knows its own installed rate by phase, from its own finished jobs, is bidding with information its competitors are guessing at. That is worth more than any single job’s recovery.

The inefficiency argument. Extended-duration and stacking positions generally depend on a comparison: here is the rate we achieved in the area we had to ourselves, here is the rate in the area we shared with three trades, here are the logs that say why. That is a quantified position. “We lost time” is not. (How a claim like this is presented and preserved depends on your contract and your state, so take it to your attorney; this is not legal advice.)

This week’s decision. Pull two people off and put them on the job that is going well. Split the crew. Call the superintendent before the area gets worse. All three need a number by Friday, not a variance report in April.

How SubMark handles it

Be clear about what SubMark does and does not do here, because the honest division of labor is the whole point.

What it gives you is the denominator, reliably.

• Each hour is recorded once, from the time clock or the daily log’s labor entries, per worker, per job, per day — not as a headcount. Field users are free and unlimited on every plan, so nobody gets left off the log to save a seat.
• Hours are costed at the rate that applied on that date, so a raise this month does not rewrite last month’s ratio.
• Labor is broken out by phase or work type when phase tracking is on — shipped on for the drywall pack, a switch in Settings for every other trade — and the phase names come from your trade’s pack, so the breakdown reads in your own words rather than in generic divisions.
• Job costing runs per job against the budget lines you wrote, grouped by cost type, with estimated, actual and variance on each, and overrun alerts on the job total and on labor specifically.
• T&M tag hours are a separate record. Hours written on a T&M tag (an opt-in module) do not post into job-costing actuals. The same crew’s clocked or logged hours for that day still do, so taking extra work out of the base-scope denominator — rule two above — is still a step you take yourself.

What it does not give you is the numerator. There is no installed-quantity field and no units-per-hour report. Job costing is in dollars and hours at the wage rate, with no burden added — work out your loaded rate yourself; see calculating a burden rate.

So the quantity stays yours. In practice that means two things. The foreman puts the number in the daily log’s work summary as a number rather than a description — “set 46 devices, levels 3 and 4 east” instead of “worked on devices” — and the division happens in a one-tab sheet in your office, weekly, per phase. That is a five-minute job once the hours are clean, and it is impossible while they are not.

What to do this month

1. Pick one unit for each of your three main phases. Something a foreman can count from where he is standing.
2. Go back to one finished job and calculate the actual rate per phase. Compare it to what the estimate implied. Do not skip this because the data is messy; a rough number from a real job beats a clean number from a rate book.
3. Add the quantity to the daily log as a number on the work summary, for one job, for four weeks.
4. Make sure extra work and back-charge hours have somewhere to go that is not the install code. If they do not, your ratio will be wrong in the flattering direction. Start with your cost code structure.
5. Run the ratio once a week, by area, and bring it to the same meeting that reviews the cost report. The first week it will tell you nothing. By the fourth it will be the first number anybody looks at.

A labor cost report tells you how much of the money is gone. Units per hour tells you whether the work went with it — and that is the difference between finding out in week six and finding out at closeout.